Miloš Savić

The structure and evolution of scientific collaboration in Serbian mathematical journals

Digital preservation of scientific papers enables their wider accessibility, but also provides a valuable source of information that can be used in a longitudinal scientometric study. The Electronic Library of the Mathematical Institute of the Serbian Academy of Sciences and Arts (eLib) digitizes the most prominent mathematical journals printed in Serbia. In this paper, we study a co-authorship network which represents collaborations among authors who published their papers in the eLib journals in an 80 year period (from 1932 to 2011). Such study enables us to identify patterns and long-term trends in scientific collaborations that are characteristic for a community which mainly consists of Serbian (Yugoslav) mathematicians. Analysis of connected components of the network reveals a topological diversity in the network structure: the network contains a large number of components whose sizes obey a power-law, the majority of components are isolated authors or small trivial components, but there is also a small number of relatively large, non-trivial components of connected authors. Our evolutionary analysis shows that the evolution of the network can be divided into six periods that are characterized by different intensity and type of collaborative behavior among eLib authors. Analysis of author metrics shows that betweenness centrality is a better indicator of author productivity and long-term presence in the eLib journals than degree centrality. Moreover, the strength of correlation between productivity metrics and betweenness centrality increases as the network evolves suggesting that even more stronger correlation can be expected in the future.

A language-independent approach to the extraction of dependencies between source code entities

Context: Software networks are directed graphs of static dependencies between source code entities (functions, classes, modules, etc.). These structures can be used to investigate the complexity and evolution of large-scale software systems and to compute metrics associated with software design. The extraction of software networks is also the first step in reverse engineering activities. Objective: The aim of this paper is to present SNEIPL, a novel approach to the extraction of software networks that is based on a language-independent, enriched concrete syntax tree representation of the source code. Method: The applicability of the approach is demonstrated by the extraction of software networks representing real-world, medium to large software systems written in different languages which belong to different programming paradigms. To investigate the completeness and correctness of the approach, class collaboration networks (CCNs) extracted from real-world Java software systems are compared to CCNs obtained by other tools. Namely, we used Dependency Finder which extracts entity-level dependencies from Java bytecode, and Doxygen which realizes language-independent fuzzy parsing approach to dependency extraction. We also compared SNEIPL to fact extractors present in language-independent reverse engineering tools. Results: Our approach to dependency extraction is validated on six real-world medium to large-scale software systems written in Java, Modula-2, and Delphi. The results of the comparative analysis involving ten Java software systems show that the networks formed by SNEIPL are highly similar to those formed by Dependency Finder and more precise than the comparable networks formed with the help of Doxygen. Regarding the comparison with language-independent reverse engineering tools, SNEIPL provides both language-independent extraction and representation of fact bases. Conclusion: SNEIPL is a language-independent extractor of software networks and consequently enables language-independent network-based analysis of software systems, computation of design software metrics, and extraction of fact bases for reverse engineering activities.

CHARACTERISTICS OF CLASS COLLABORATION NETWORKS IN LARGE JAVA SOFTWARE PROJECTS

Understanding software structural complexity and evolution plays an important role in controlling the software development and maintenance process. Recent studies have shown that the theory behind complex networks, especially the theory of scale-free networks, can be a useful approach to the analysis of concrete software systems. In this paper, class collaboration networks associated with five large Java software systems (JDK, Ant, Tomcat, Lucene and JavaCC) are analyzed in order to determine whether they belong to the class of scale-free networks, and examine their small-world characteristics. For each analyzed network, we detected (approximately) scale-free and (ultra) small-world properties. The results indicate that general conclusions from scale-free network theory can be applied to Java software systems in order to understand their structural complexity and model software evolution at the structural (class collaboration) level. Moreover, we examine class collaboration network evolution of Ant, in order to check the preferential attachment hypothesis of the Barabasi-Albert model. For several major Ant network transitions we con-clude that preferential attachment can successfully model Ant evolution at the class collaboration level. Finally, we discuss the implications of our results on software engineering, in several aspects: identification of important clas-ses/interfaces, software testing strategy, and efficient communication among software entities. http://dx.doi.org/10.5755/j01.itc.40.1.192

Language independent framework for static code analysis

The aim of this paper is to describe a framework consisting of a set of static analyzers. The main characteristic of all incorporated tools is their independency of input programming language. This independency is based on a universal representation of the source code that is to be analyzed. The overall goal of this research is to provide a framework that is suitable for consistent analysis of the source code with the intention to ensure, check, and consequently increase the quality of the heterogeneous software products. The framework currently integrates three tools: software metrics tool -- SMIILE, extractor of software networks -- SNEIPL and structure change analyzer -- SSCA, with tendency to extend this set of components. Further application of these tools in collaboration with other tools on higher level provides even broader applicability of described framework.

Exploratory Analysis of Communities in Co-authorship Networks: A Case Study

Digital libraries enable worldwide access to scientific results, but also provide a valuable source of information that can be used to investigate patterns and trends in scientific collaboration. The Electronic Library of the Mathematical Institute of the Serbian Academy of Sciences and Arts (eLib) digitizes the most prominent mathematical journals printed in Serbia. Using eLib bibliographical records we constructed a co-authorship network representing collaborations between authors who published their papers in eLib journals in the period from 1932 to 2011. In this paper we apply community detection techniques in order to examine the structure of the eLib co-authorship network. Such study reveals characteristic patterns of scientific collaboration in Serbian mathematical journals, and helps us to understand the (self-)organization of the eLib community of authors.

Extractor of software networks from enriched concrete syntax trees

Software networks are directed graphs of dependencies among entities (packages, classes, methods, etc.) found in a software system. These structures are used to study organizational software complexity and evolution. In this paper the prototype of software networks extractor - SNEIPL is described. SNEIPL uses enriched concrete syntax tree (eCST) representation of a source code. The benefit of such representation is that it enables independency of programming language. First experiences using SNEIPL showed that it correctly extracted isomorphic software networks from the source code of two structurally and semantically equivalent programs written in different programming languages.

Analysis of intra-institutional research collaboration: a case of a Serbian faculty of sciences

Current research information systems (CRISs) offer great opportunities for scientometric studies of institutional research outputs. However, many of these opportunities have not been explored in depth, especially for the analysis of intra-institutional research collaboration. In this paper, we propose a hybrid methodology to analyze research collaboration networks with an underlying institutional structure. The co-authorship network extracted from the institutional CRIS of the Faculty of Sciences, University of Novi Sad, Serbia, is analyzed using the proposed methodology. The obtained results show that the organizational structure of the institution has a profound impact on both inter- and intra-institutional research collaboration. Moreover, researchers involved in inter-department collaborations tend to be drastically more productive (by all considered productivity measures), collaborative (measured by the number of co-authorship relations) and institutionally important (in terms of the betweenness centrality in the co-authorship network) compared to those who collaborate only with colleagues from their own research departments. Finally, our results indicate that quantifying research productivity by the normal counting scheme and Serbian research competency index is biased towards researchers from physics and chemistry research departments.

SSQSA architecture

The aim of this paper is to describe architecture of the software system called Set of Software Quality Static Analyzers (SSQSA). The main aim of SSQSA is to provide some static software analyzers to ensure, check, and consequently increase, the quality of software products. Its main characteristic is the language independency which makes it more usable than many other similar systems.

Community detection and analysis of community evolution in Apache Ant class collaboration networks

In this paper we investigate community detection algorithms applied to class collaboration networks (CCNs) that represent class dependencies of 21 consecutive versions of the Apache Ant software system. Four community detection techniques, Girvan-Newman (GN), Greedy Modularity Optimization (GMO), Walktrap and Label Propagation (LP), are used to compute community partitions. Obtained community structures are evaluated using community quality metrics (inter- and intra-cluster density, conductance and expansion) and compared to package structures of analyzed software. In order to investigate evolutionary stability of community detection methods, we designed an algorithm for tracking evolving communities. For LP and GMO, algorithms that produce partitions with higher values of normalized modularity score compared to GN and Walktrap, we noticed an evolutionary degeneracy -- LP and GMO are extremely sensitive to small evolutionary changes in CCN structure. Walktrap shows the best performance considering community quality, evolutionary stability and comparison with actual class groupings into packages. Coarse-grained descriptions (CGD) of CCNs are constructed from Walktrap partitions and analyzed. Results suggest that CCNs have modular structure that cannot be considered as hierarchical, due to the existence of large strongly connected components in CGDs.

A Feature Selection Method Based on Feature Correlation Networks

Feature selection is an important data preprocessing step in data mining and machine learning tasks, especially in the case of high dimensional data. In this paper we present a novel feature selection method based on complex weighted networks describing the strongest correlations among features. The method relies on community detection techniques to identify cohesive groups of features. A subset of features exhibiting a strong association with the class feature is selected from each identified community of features taking into account the size of and connections within the community. The proposed method is evaluated on a high dimensional dataset containing signaling protein features related to the diagnosis of Alzheimer’s disease. We compared the performance of seven widely used classifiers that were trained without feature selection, with correlation-based feature selection by a state-of-the-art method provided by the WEKA tool, and with feature selection by four variants of our method determined by four different community detection techniques. The results of the evaluation indicate that our method improves the classification accuracy of several classification models while drastically reducing the dimensionality of the dataset. Additionally, one variant of our method outperforms the correlation-based feature selection method implemented in WEKA.